AI can potentially help with medical research, learning aids, and possibly robots that could take over the earth according to several movies. But in their ambition to win the AI race at all costs, major tech companies are causing significant harms now, starting with their choice to backtrack on their renewable energy and climate commitments by powering data centers with fossil fuels.

AI data centers nationwide are extending the life of fossil fuels – natural gas and coal plants -- polluting communities and making climate change worse. Alarmingly, new natural gas plants are being built all over the country to power AI. There are plans for over 100 new natural gas plants in Texas alone, driven in large part by AI.

One of the scary parts of the fossil fuel building energy boom is that once fossil fuel plants are restarted or built by mostly regional utilities, those companies aren’t likely to just abandon them all. Once these plants are built, we will be stuck with them (known as stranded assets) and their pollution for generations whether all the AI speculation comes to fruition or not.

And, even if some fossil fuel plants shut down, the cost of building those plants will be passed onto ratepayers like us for years to come. The ironic truth is these Big Tech companies know that renewable energy combined with increased efficiency can meet their needs, will help the world reach climate goals, and is cheaper to build and deploy.

Yes! Experts estimate that by 2030, AI energy use will equal the amount of energy currently needed to power 22% of households. Of course, Big Tech companies are not transparent around their growing energy needs, and some of the data centers they are planning may never get built, but the surge in demand for energy is still going to be huge.

In the end, it will be us, the everyday consumers, and our communities. In states where data centers are being built, households are seeing major increases in their monthly electricity bills. For example, rate hikes in Oregon are up 50% and Georgia bills are about $500 annually.

A recent analysis from Bloomberg found that in areas with extensive data centers, household energy bills for one month cost a whopping 267% more than five years ago. Bottom line: The cost for any new energy infrastructure ends up in household energy bills. And IF the AI energy speculation turns out to be less than expected, communities will still be stuck with the sunk costs of infrastructure, even if it is not used.

Fossil fuels are the most polluting form of energy, bar none. Proven issues with air pollution, water pollution, climate change, and mining issues are well documented. This is an especially critical issue of justice for marginalized communities, as they frequently bear the greatest burdens from fossil fuel extraction and burning.

While running a nuclear power plant doesn’t produce climate emissions, nuclear power is anything but clean. From the toxic impacts of uranium mining to nuclear waste that remains radioactive for decades, nuclear power is dangerous. Besides restarting 3-mile Island, the site of the worst nuclear disaster in the US, tech companies are looking to power their data centers by building small modular reactors (SMR) that proponents claim are cheaper and safer. But by most accounts, this unproven “solution” may be years away from implementation, and still doesn’t address the concerning issues with nuclear power.

Water to cool the data centers brings oceans of problems. Data centers are being built in areas already struggling with water scarcity. Some communities are even finding that their taps are running dry thanks to AI.

Proponents of data centers like to hype job creation, tax revenues and infrastructure upgrades that data centers can bring, and many local politicians fall in line because they want to be seen as strengthening the local economy. But many of the jobs created are temporary, companies are often offered tax incentives to open data centers (offsetting revenues), and infrastructure enhancements pale in comparison to the noise and air pollution that they produce, as well as increased energy bills in the community. And these harms often occur in marginalized communities that are often already impacted by pollutants.

Without extensive community input, and the ability of local residents and businesses to have a say over how and where data centers are built and how they are powered, the costs of data centers will continue to outweigh the benefits in most places.

The term AI itself is a misnomer, it’s neither self-aware nor possesses the ability to think and feel. What AI is really doing is a search of the internet on steroids, one that returns results based on information popularity, not accuracy. It is learning at a greater rate than yesterday’s Google search, but that sophistication comes with a price too. Intellectual property rights are being circumvented leaving artists, writers, and other creatives without the usual protections and compensation for their work. AI is increasingly used to help fight wars. AI also perpetuates biases when used in surveillance, exacerbating existing patterns of racism. And by surfacing information based on what gets the most engagement rather than what’s most accurate and contextual, AI search can make it easier to spread mis- and disinformation online.

Tech companies are quick to hide behind claims of intellectual property rights (IPR), which is ironic given how many tech companies have ignored IPR of artists, writers, and other creators in training their AI models! Companies like Google, Meta, Amazon, and Microsoft are claiming their development of AI must be kept secret to protect their business models. Sure, to be good neighbors and not leave communities with stranded assets, these companies must be more open about the resources from water to electricity that will be required to run the data centers that will then not be available to communities.